Increasing net.ipv4.ip_local_port_range beyond the default WSL range causes localhost TCP connections to hang

Nobody has claimed this yet.

### Windows Version

Microsoft Windows [Version 10.0.26200.8875]

### WSL Version

2.6.3.0

### Are you using WSL 1 or WSL 2?

- [x] WSL 2
- [ ] WSL 1

### Kernel Version

_No response_

### Distro Version

_No response_

### Other Software

_No response_

### Repro Steps

Changing net.ipv4.ip_local_port_range from the WSL default causes all localhost TCP connections to hang. This affects multiple applications (Redis, Python HTTP server, curl, nc) and is not specific to a single service.

The issue is reproducible and is resolved only after running wsl --shutdown.

1. Start a fresh WSL session.

2. Verify localhost connectivity.

$ redis-cli ping

PONG

3. Verify the default ephemeral port range.

$ cat /proc/sys/net/ipv4/ip_local_port_range

44620 48715

4. Change the ephemeral port range.

sudo sysctl -w net.ipv4.ip_local_port_range="44620 65535"

(or)

sudo sysctl -w net.ipv4.ip_local_port_range="32768 60999"

5. Test localhost connectivity again.

redis-cli ping

or

python3 -m http.server 8000

curl http://127.0.0.1:8000

The attached screenshot shows:

redis-cli ping and curl http://127.0.0.1:8000 working with the default ephemeral port range (44620 48715).
Both commands hanging after changing net.ipv4.ip_local_port_range to 10000 65535 or 32768 60999.

### Expected Behavior

Changing the ephemeral port range should not affect localhost TCP connectivity.

### Actual Behavior

redis-cli ping hangs indefinitely.
curl http://127.0.0.1:8000 hangs.
nc hangs.
ss -tanp shows connections remaining in SYN-RECV.
redis-server continues listening on 127.0.0.1:6379.
Running wsl --shutdown restores localhost connectivity.

### Diagnostic Logs

Diagnostic information

While redis-cli is hanging:

$ ss -tanp | grep 6379

LISTEN 127.0.0.1:6379
SYN-RECV 127.0.0.1:6379 127.0.0.1:xxxxx

Start by reproducing the WSL 2 behavior with /proc/sys/net/ipv4/ip_local_port_range, sysctl, redis-cli, curl, and ss -tanp. Compare localhost connectivity before and after changing the range, and use wsl --shutdown to confirm the reported reset; done means localhost TCP connections continue working after the change without requiring shutdown.

Written by the indexing model from the issue text.